Claude Code v2.1.197: Claude Sonnet 5 Default, 1M Tokens, Promo Pricing
Claude Code v2.1.197 ships with Claude Sonnet 5 as the default model. The new model offers a native 1M-token context window and promotional pricing of $2 per million input tokens and $10 per million output tokens through August 31. Update to version 2.1.197 to get access.
Key Details
- Default Model Change: Claude Sonnet 5 is now the default in Claude Code. No manual model selection required after updating.
- Context Window: Native 1M-token context — same size as Claude 3.5 Sonnet, but now built in without chunking.
- Pricing Promotion: $2/Mtok input, $10/Mtok output until August 31. After the promo, pricing reverts to standard rates (not specified in source).
- Version Requirement: Must be on v2.1.197 to use Claude Sonnet 5 in Claude Code.
No other changes or fixes are documented in this release. The update is available now for all Claude Code users. To upgrade, run the standard update command for your setup.
📖 Read the full source: GitHub Claude-Code
👀 See Also
AI Didn't Delete Your Database — You Did: Accountability in the Age of AI Coding Agents
A viral story blamed an AI agent for deleting a production database, but the real issue is exposing destructive API endpoints and lack of process—not the tool.
AI Scans 400K Reddit Posts, Uncovers Hidden Ozempic Side Effects Like Menstrual Changes
University of Pennsylvania researchers used LLMs to analyze 400,000 Reddit posts on GLP-1 drugs, uncovering underreported symptoms like menstrual irregularities and chills.
llama.cpp Q8_0 quantization gets 3.1x speedup on Intel Arc GPUs with SYCL reorder fix
A fix to llama.cpp's SYCL backend brings Q8_0 quantization on Intel Arc GPUs from 21% to 66% of theoretical memory bandwidth, achieving 15.24 tokens/second versus 4.88 tokens/second previously on an Arc Pro B70 with Qwen3.5-27B.
Claude Opus 4.6 accuracy drops on BridgeBench hallucination test
Claude Opus 4.6 shows a significant drop in accuracy on the BridgeBench hallucination test, falling from 83% to 68% according to BridgeMind AI's Twitter post.